ANNUAL JOINT WMO TECHNICAL PROGRESS REPORT ON
THE GLOBAL DATA PROCESSING AND FORECASTING SYSTEM (GDPFS) INCLUDING NUMERICAL WEATHER PREDICTION (NWP) RESEARCH ACTIVITIES IN 2008
China, May 2009
1. Summary of highlights
1.1 Developments of operational NWP
· The Global Medium-range forecast system TL639L60 was put into operational use on 1 June 2008.
· The GRAPES_Meso 15Km system replaced GRAPES_Meso 30Km system for distributing the products in April 2008.
· The new-generation global TC track numerical prediction system began its operational run in July 2008.
Developments of GRAPES
· A new global forecast system--GRAPES_GFS with 50km horizontal resolution and 36 vertical levels has been developed, and 1-year assimilation-forecast cycle experiments have been accomplished. This system would be put into quasi-operational test in the early 2009 at NMC/CMA.
· A high-order positive-definite scalar advection scheme has been implemented in GRAPES_Meso system, and substantive improvement is obtained in the precipitation forecast, especially for the heavy rainfall forecast.
2. Equipment in use at the Centre
· The mainframe computer for numerical analysis and prediction in CMA was still IBM CLUSTER 1600 high-performance parallel computer system, which was introduced in CMA in 2004.
· Beijing-Bangkok circuit was upgraded from X.25 link to TCP/IP link with the transmission speed of 64kbps in January 2008.
· The backup connection between CMA and EUMETSAT was established via Internet in 2008. It was operating in parallel with CMA-EUMETSAT connection to exchange the observations and products of FY-2C, METEOSAT 7, METEOSAT 9, GOES 11 and 12 on a near real-time basis.
3. Data and Products from GTS in use
No changes
4. Forecasting system
4.1 System run schedule and forecast ranges
eneral structure of a prognostic system, models in operational use, run schedule, forecast ranges]"
· The new global model forecast system (T639) with the global data assimilation (GSI) has run in quasi-operational since the end of 2007. The products of the system were disseminated nationwide as from 1 June 2008. .It produces global analyses (including data assimilation) at the 4 main synoptic hours 00, 06, 12 and 18 UTC and global 10-day forecasts at 00 and 12 UTC on a routine basis.
· The Typhoon relocation scheme has been used in the operational TC track forecast system. It was used in the continuous tests for making TC ensemble forecasts for the synoptic hours and TC forecasts at 00 and 12 UTC when a typhoon occurs.
Operating Schedule of NWP Systems at NMC
(GMT) / Wall clock
(GMT) / Computer
Global Model Forecast System
(T639L60_GSI) / 04:00 (18Z_ASSIM+9HR_FCST) / 04:00～01:15 / IBM Cluster 1600
05:40 (00Z_ASSIM+240HR_FCST) / 05:40～08:30 / IBM Cluster 1600
12:00 (00Z_ASSIM+9HR_FCST) / 12:00～13:15 / IBM Cluster 1600
16:10 (06Z_ASSIM+9HR_FCST) / 16:10～17:25 / IBM Cluster 1600
17:40 (12Z_ASSIM+240HR_FCST) / 17:40～20:30 / IBM Cluster 1600
23:00 (12Z_ASSIM+9HR_FCST) / 23:00～00:15 / IBM Cluster 1600
03:30 (00Z_ASSIM+120HR_FCST) / 03:35～04:00 / IBM Cluster 1600
Global Model / 06:30 (00Z_ASSIM.） / 06:35～06:40 / IBM Cluster 1600
(T213L31) / 10:30 (06Z_ ASSIM +72HR_.FCST) / 10:35～10:45 / IBM Cluster 1600
16:30 (12Z_ASSIM.+240HR_FCST) / 16:35～17:15 / IBM Cluster 1600
22:10 (18Z_ASSIM.+72HR_FCST) / 22:15～22:25 / IBM Cluster 1600
05:00 (00Z_ ASSIM +240HR_FCST) / 05:00～05:40 / IBM Cluster 1600
Typhoon Track Forecast System
(T213L31_SSI) / 11:00 (00Z_ASSIM. +9HR_FCST) / 11:00～11:15 / IBM Cluster 1600
12:00 (06Z_ASSIM.+120 HR_FCST) / 12:00～12:30 / IBM Cluster 1600
16:00 (06Z_ASSIM.+9HR_FCST) / 16:00～16:15 / IBM Cluster 1600
17:00 (12Z_ASSIM. +240HR_FCST) / 17:00～17:40 / IBM Cluster 1600
22:00 (12Z_ASSIM. +9 HR_FCST) / 22:00～22:15 / IBM Cluster 1600
22:30 (18Z_ASSIM. +120 HR_FCST) / 22:30～23:00 / IBM Cluster 1600
04:00 (18Z_ASSIM. +9 HR_FCST) / 04:00～04:15 / IBM Cluster 1600
Regional Model (MM5V3) / 03:00(00Z+48HR_FCST) / 03:05～03:35 / IBM Cluster 1600
16:00 (00Z+48HR_FCST) / 16:05～16:35 / IBM Cluster 1600
Regional Model (GRAPES) / 04:15 (00Z_ ASSIM +60HR_.FCST) / 04:20～04:40 / IBM Cluster 1600
13:40 (06Z_ ASSIM) / 13:45～13:55 / IBM Cluster 1600
16:50 (00Z_ ASSIM +60HR_.FCST) / 16:55～17:15 / IBM Cluster 1600
22:30 (18Z_ ASSIM) / 22:35～22:40 / IBM Cluster 1600
Ensemble Prediction
15 members
(T213L31) / 07:30 (00Z_ASSIM+6HR_FCST) / 07:30～07:35 / IBM Cluster 1600
12:30 (06Z_ASSIM+6HR_FCST) / 12:30～12:35 / IBM Cluster 1600
18:30 (12Z_ASSIM+240HR_FCST) / 18:30～20:15 / IBM Cluster 1600
23:30(18Z_ASSIM+6HR_FCST) / 23:30～23:35 / IBM Cluster 1600
Prediction
15 members
(WRF) Ensemble / 07:00(36HR_FCST) / 07:00～07:40 / IBM Cluster 1600
14:00 (00Z_ASSIM+6HR_FCST) / 14:00～14:15 / IBM Cluster 1600
14:15 (06Z_ASSIM+6HR_FCST) / 14:15～14:30 / IBM Cluster 1600
19:00 (36HR_FCST) / 19:00～19:40 / IBM Cluster 1600
00:00 (12Z_ASSIM+6HR_FCST) / 00:00～00:15 / IBM Cluster 1600
00:15 (18Z_ASSIM+6HR_FCST)) / 00:15～00:30 / IBM Cluster 1600
Typhoon Track Ensemble Prediction
15members
(T213L31+Bogus) / 07:30 (00Z_120HR_FCST) / 07:30～08:30 / IBM Cluster 1600
19:30 (12Z_120HR_FCST) / 19:30～20:30 / IBM Cluster 1600
Sand Storm (MM5) / 05:30 (72HR_FCST) / 05:30～06:30 / IBM Cluster 1600
18:30 (72HR_FCST) / 18:30～19:30 / IBM Cluster 1600
Wave Watch (WW3) / 07:00 (120HR_FCST) / 07:00～07:20 / IBM Cluster 1600
19:00 (120HR_FCST) / 19:00～19:20 / IBM Cluster 1600
4.2 Medium range forecasting system (4-10 days)
4.2.1 Data assimilation, objective analysis and initialization
4.2.1.1 In operation
nformation on Data assimilation, objective analysis and initialization]"
· The data assimilation system used in operation is Grid-point Statistical Interpolation (GSI) which was introduced from NECP. Compared with T213_OI NWP system, the T639_GSI can improve the north hemisphere by about 1 day in lead time. At present, the conventional observational data from GTS and the NOAA-15/16/17 satellites ATOVS 1b data are assimilated in the system. But some new satellite data, such as ATOVS NOAA-18 and METOP-2 radiance observations will also be assimilated in the system soon.
4.2.1.2 Research performed in this field
· The GRAPES global 3DVAR system (GRAPES_GAS) is an incremental grid-point data analysis system with 1.125˚ x 1.125˚ horizontal resolution and 17 standard pressure levels. The data assimilated include the conventional GTS data, NOAA 15, 16, 17 and 18 radiances, METEOSAT9 and MTSAT AMV. The analysed variables include zonal and meridianal winds, geopotential height and relative humidity, and the first guess is from GRAPES global 6-hour forecast with the digital filter.
· The development of GRAPES global tangent linear and adjoint model has been accomplished. The parallelization of the code is under way in preparation for a 4-D VAR system.
4.2.2 Model
4.2.2.1 In operation
· TL639L60 began its operational run from 1 June 2008 with a horizontal resolution of T639 (30 km) and 60 vertical levels (up to 0.1 hPa)，and time step is 600s. 10-day forecasting is made twice a day. It is a hydrostatic model, and its physics Include orography (mean topography), gravity wave drag, 4 surface and sub-surface levels (allowing for vegetation cover, gravitational drainage, capillarity exchange, surface and sub-surface runoff, deep-layer soil temperature and moisture); evaporation, sensible and latent heat flux; planetary boundary layer, horizontal diffusion, radiation (incoming short-wave and out-going long-wave); advective and convective precipitation, snow-fall, etc.
4.2.2.2 Research performed in this field
ummary of research and development efforts in the area]"
· The GRAPES global model (GRAPES_GAM) is a non-hydrostatic grid-point model with 50km horizontal resolution and 36 vertical levels. The structure of GRAPES_GAM is described below:
Ø Equations: Fully compressible equations with shallow atmosphere approximation
Ø Variables: Zonal wind U, meridianal wind V, vertical velocity W, potential temperature θ, specific humidity q(n) and Exner function Π.
Ø Numerical technique: 2-time level semi-implicit semi-Lagrangian time-stepping method; 3D vector formulation for the momentum equation; Quasi-monotone positive-definite scalar advection; non-hydrostatic.
Ø Horizontal grid: Modified Arawaka C-grid with V-point at the poles.
Ø Time step: 600s
Ø Vertical grid: Terrain-following Z-coordinate with Charney-Phillipps variable distribution
Ø Physics: Subgrid scale orography-induced gravity wave drag, RRTM LW / Fouquart & Bonnel SW, simplified Arakawa-Schubert cumulus, NCEP-cloud 5 microphysics, MRF vertical diffusion, simple land surface process with energy balance and thermal diffusion.
Ø Boundary forcing: Nearly real-time SST, sea-ice climatology.
· The Yin-Yang grid is being implemented into GRAPES_GAM in replacement of the lat-lon grid. And, the terrain-following Z coordinates are replaced by the hybrid coordinates (the terrain-following Z plus Z coordinates). The CoLM is being incorporated into GRAPES_GAM.
4.2.3 Operationally available NWP Products
rief description of variables which are outputs from the model integration]"
variables / unit / Forecast hours / Levels （hpa） / AreaTemperature / K / 00,03,06, 09,12,15,18,21,24,27,30,33,
36,39,42,45,48,51,54,57,60,
63,66,69,72,75,78,81,
84,87,90,93,96,99,102,105
108,111,114,117,120,126,132,138, 144,150,156,162
168,180, 192,204, 216,228
240 / 10,20, 30,50,
70,100, 150,200,
250,300, 400,500,
600,700, 850,925,
1000 / Global grid(0.28125*0.28125)
Height / M
U-wind / m/s
V-wind / m/s
W / pa/s
q / Kg/kg / 50,70,100, 150,200, 250,300,
400,500, 600,700,
850,925, 1000
rh / %
Sea level pressure / hpa / Sea level
vorticity / s-1 / 00,03,06,
09,12,15,18,21,24,27,30,33,
36,39,42,45,48,51,54,57,60,
63,66,69,72,75,78,81,
84,87,90,93,96,99,102,105
108,111,114,117,120,126,132,138, 144,150,156,162
168,180, 192,204, 216,228
240 / 10,20,30,50,70,
100,150,200,250,
300,400,500,600,
700,850,925,1000 / Global grid(0.28125*0.28125)
divergence / s-1
Temperature / K / Surface
Pressure / hpa / Surface
10mU / m/s / 9999
10mV / m/s / 9999
2mT / K / 9999
2mRH / % / 9999
Soil temperature / K / 1,2,3,4
Soil moisture / kg/m2 / 1,2,3,4
Rainfall / mm / 9999
Temperature Advection / 10-6C/s / 200,500,700,850
925,1000 / North-east hemisphere
Vorticity advection / 10-11/s2
T-TD / 10-1°C / 00,06,12,18,24,30,36,42,48,
Vapor Flux / 10-1g/
hap×cm×s
10-7g/
hap×cm²s / 60,72, 96,
Vapor Flux divergence / 120,144, 168
qse / K
K index / °C
4.2.4 Operational techniques for application of NWP products (MOS, PPM, KF, Expert Systems, etc..)
4.2.4.1 In operation
rief description of automated (formalized) procedures in use for interpretation of NWP ouput]"
· In routine weather element forecast, predictions of low cloud cover, total cloud cover, daily mean wind speed and daily mean temperature were added to routine operation by using MOS technique.
4.2.4.2 Research performed in this field
· An experiment on objective temperature forecast was conducted to introduce temperature observations as a factor of temperature forecasting equation. In order to contrast with traditional MOS, a control test was made to predict temperature without introducing temperature observations. The 662 test stations were selected in comparison with MOS outputs. For 48, 72, 96, 120, 144, 168, 192, 216, 240 264 hours, the control forecast by traditional MOS technology was better than the forecast under test. For 24 hour maximum temperature prediction, the test forecast was better than control test except Northeast, Southwest distract. Taking into account all stations as a whole, the mean skill of test forecast was 67.7%, and that of the MOS forecast was 67.0%. For 24-hour minimum temperature prediction, the MOS forecast was better than control test. On a whole, the average accuracy in minimum temperature within a range of 2°C in test forecast was 77.1%, and that for the MOS forecast was 76.6%.
· In objective wind predictions, wind was predicted by kernel neighbor nonparametric estimation technique from 1 December 2007 to 30 November 2008. More than 2200 stations in China were tested in 7 days with 56 forecasts in wind direction and speed. A control test was compared with DMO (Direct Model outputs) in the same period. In cases of winter, summer and fall, the test predictions were better than DMO control test in wind direction, while the control test predictions were better than that of test in the spring. For wind speed prediction of less than 3.3m/s, the mean TS skill of test prediction was 0.768, while that of DMO test was 0.752. For wind of more than 3.4m/s, the TS score for the test was 0.233, for DMO forecast it was 0.154. For wind beyond 8.0m/s, TS score for test forecast was 0.139, and 0.047 for DMO forecast. The new method was found as a useful tool for wind prediction.
4.2.5 Ensemble Prediction System (EPS)
4.2.5.1 In operation
umber of runs, initial state perturbation method, perturbation of physics?]" (Describe also: time range, number of members and number of models used: their resolution, number of levels, main physics used )
· An updated global Ensemble Prediction System was put into quasi-operational use in June 2008: The data assimilation system for control forecast was upgraded from the optimal interpolation analysis system to the 3-D VAR assimilation system, with the capabilityto directly assimilate ATOVS data.
The system configuration is as follows:
Ø Number of members: 15-members; 14 perturbed members (adding/subtracting perturbations from 7 independent breeding cycles) plus control run.
Ø Initial state perturbation method: Breeding method
Ø Number of models used: 1 model, T213lL31
Ø Perturbation of physics: No
Ø Running cycle : 12Z running each day
- T213L31 resolution control out to 10 days.
- 14 perturbed forecasts each run at T213L31 resolution and out to 10 days.
- The perturbations are from seven independent breeding cycles.
4.2.5.2 Research performed in this field
ummary of research and development efforts in the area]"
4.2.5.3 Operationally available EPS Products
rief description of variables which are outputs from the EPS"
l Post-processing:
Mean/spread: Hgts (mean) and vorticity (spread) at 250hPa, 500hPa;
relative humility at 700hPa, 850hPa;
temperature at 850hPa, 700hPa,500hPa,250 hPa;
wind at 850hPa, 700hPa,500hPa,250hPa;
10m wind;
2m temperature
1000-850 hPa thickness, 1000-5000 hPa thickness, 850-700 hPa thickness;
precipitation
sea level pressure
l Spaghetti：
2m temperature,
MAY-SEP: Highest (30,35), OCT-APR:Lowest (-8,5); (0,12)
850hPa temperature: MAY-SEP (12,24), or OCT-APR (-12,0,12）
500 hPa Hgts
MAY-SEP: (512,564); (568,588)
OCT-APR: (512,560);(544,588)
sea level pressure, 24-h total precipitation,
l Probability: 10m wind, 2m temperature, precipitation.
4.3 Short-range forecasting system (0-72 hrs)
4.3.1 Data assimilation, objective analysis and initialization
4.3.1.1 In operation
nformation on Data assimilation (if any), objective analysis and initialization,]" (Indicate boundary conditions used)
· The GRAPES regional 3DVAR system is an incremental grid-point data analysis system with 15km horizontal resolution and 31 vertical levels the same as the GRAPES_Meso model. The data assimilated include the conventional GTS data, NOAA 15, 16 and 17 radiances. The analysed variables include zonal and meridonal winds, geopotential height and relative humidity, and the first guess is from the operational T213 global model 6-hour forecast with the digital filter as the initialization.